FIX: plastik style poor color choices for tree views

[Matthew Woehlke]

Remi Villatel wrote:
> On Monday 05 February 2007 16:15, Matthew Woehlke wrote:
>>> In certain color schemes (particularly dark window background color),
>>> the Plastik style makes poor color choices that can make the lines in
>>> tree views, and the boxes around the expand icons, nearly invisible.
> 
> If this was the only problem...
> [snip]
> Plastik is almost unusable if your use white text on a very dark background, 
> a lot of framings become invisible.

Well, I'm not using white-on-black, I'm using light-on-dark, in which 
case, the tree view is the only problem I recall noticing (I don't 
actively use Plastik). White-on-black causes problems for *most* styles. 
IMO if you want to use actual-white on actual-black, use a style that 
works well with that.

The problems I was addressing occur when 1: using dark-but-not-black 
backgrounds, and 2: when mid() and base() are the same or similar. (1) 
may be less common, but (2) is not; e.g. the "CDE", "Digital CDE", "Dark 
Blue" and "Solaris" schemes all would exhibit this problem.

> To support dark colorschemes, my style (Serenity) makes a really extensive 
> use of its equivalent of alphaBlendColors(), especially to replace any 
> fixed .light() and .dark() which have the inverse of the appropriate effect 
> when the colors are inverted. background.dark() for example doesn't produce 
> any visible result when background == #000000, neither does 
> foreground.light() when foreground == #FFFFFF.

Hmm... this brings to mind the time a while back when I asked for 
alphaBlendColors-equivalent to be migrated to kdelibs, to which the 
response was 'no, Qt should do it'. Any Qt developers reading? Any 
progress, or should I go back to working on my kdelibs solution?

This is something that IMO really needs to be in place ASAP so porting 
efforts can take advantage of it.

-- 
Matthew
This message is non-smoking